Skip to content

TerraFlow

A reproducible framework for climate-impact assessment of agricultural suitability — including climate-induced crop hazards (drought, flood, heat stress, growing-degree-day shifts) under historical and projected future climate. Give it a land-cover map, a climate dataset, and a config file — it returns a scored, location-stamped results table with per-cell uncertainty intervals and a content-addressable run fingerprint that lets reviewers regenerate the exact same numbers three years later.

The same workflow methodology extends to habitat suitability, land-use planning, and conservation siting.

Choose your path

  • TerraFlow in 10 Minutes

    New to TerraFlow? Start here — no coding required. Covers what it does, the problem it solves, how the pipeline works, and what the output means.

    Get started

  • Field Guide

    You receive results but don't run TerraFlow yourself. Plain-English guide to scores, labels, opening results in Excel or QGIS, and sharing with others.

    Read the Field Guide

  • Technical Reference

    You install, configure, or contribute to TerraFlow. CLI, YAML schema, architecture docs, ADRs, and API reference.

    CLI Usage · Config Schema

What's available

Per-cell climate values
Spatial interpolation via linear, kriging, or IDW — not a single global mean.
Kriging + uncertainty
interpolation_method: kriging adds per-cell {var}_krig_std columns. Pair with uncertainty_samples for Monte Carlo score confidence intervals.
CRS-aware ROI clipping
Supply your bounding box in WGS 84 degrees regardless of the raster's native projection.
Guaranteed WGS 84 output
lat/lon columns always contain geographic degrees, safe for any mapping tool.
Reproducible sampling
Cell selection is seeded from the run fingerprint — same config always yields the same rows.
Portable configs
Relative paths resolve against the config file's location, not the caller's working directory.

Technical documentation

Start running:

Understand the system:

Contribute:

API reference: